Evidence against hyperspecificity in implicit invariant learning.

Four experiments examined the claim that cross-format transfer in invariant learning is reliant solely on the presence of repetition structure in study and test strings (Stadler, Warren, & Lesch, 2000). Experiments 1, 2, and 3 used strings with no repetitions and found significant cross-format transfer in combination with a non-significant transfer decrement--no significant difference between same- and changed-format conditions. Further investigation of the basis of the role of repetition structure revealed an emphasis on the perceptual salience of test stimuli (Experiment 4). Our results contrast with those of Stadler et al. and suggest that under the conditions we employed invariant learning is not highly sensitive to changes in the perceptual characteristics of stimuli and therefore is inaccurately described as hyperspecific. We suggest that the term hyperspecific be reserved for cases in which minor format changes result in significant performance impairments--for example, typographical effects in implicit memory.